Prediction uncertainties and inaccuracies resulting from common assumptions in modelling vibration from underground railways

Abstract

Underground railways produce significant ground-borne vibration that is reported to disturb people living or working near subways. Designers and engineers use numerical models to predict vibration levels so as to meet the increasingly strict vibration standards. These models commonly include simplifying assumptions to reduce the complexity and cost of the simulation. This paper reviews six commonly disregarded aspects of the underground railway environment and their respective effects on vibration prediction values: a second (twin) tunnel, piled foundations, track with discontinuous slabs, soil inhomogeneity, inclined soil layers, and irregular contact at the tunnel–soil interface. Results suggest that accounting for each of these simplifying assumptions can result in predictions that vary from the simplified cases by at least 5 dB and potentially up to 20 dB. This is a significant level of uncertainty and should be considered when estimating the predictive accuracy of numerical models using simplifying assumptions.